Global Trends in Marine Biodiversity from Unstructured Data

Marine biodiversity is changing in response to numerous, mounting anthropogenic impacts, and effective conservation and management in the face of these threats requires a robust understanding of spatial and temporal patterns and trends in marine biodiversity. Achieving such an understanding is possible with the vast amounts of data present in aggregated online repositories, such as the Ocean Biogeographic Information System (OBIS), however overcoming incomparability between constituent datasets, and issues of variable methodology, detectability, and effort, requires that we employ statistical methods that ensure derived trends are robust to bias from “unstructured data”. In this thesis, I explore how one of these methods, occupancy modelling, can be used to overcome issues of detectability and variable surveyor effort in OBIS data, while employing data management and analysis techniques to minimise the effects of variable methodologies. I use this combination of methods and aggregated data to assess temporal trends in the lesser-studied molluscs, expanding our understanding of molluscs in the Celtic Sea, and assessing the utility of multispecies models on a global scale for the genus Conus. I then go on to address more fundamental macroecological questions by deriving inter- and intraspecific abundance-occupancy relationships in European cetaceans, to then demonstrate how occupancy modelled unstructured data can be used to robustly estimate relative abundance of species within this group. Finally, I apply occupancy modelling to an Atlantic wide dataset of marine fauna, to attempt to address recent debate surrounding the marine latitudinal diversity gradient. I find throughout this thesis that occupancy modelling and unstructured data are useful in determining robust but coarse scale trends when sufficient data are available, and end by suggesting future avenues of research to both further test the methodology, and improve our knowledge of changes in marine biodiversity.